Use when executing a written implementation plan - guides step-by-step TDD execution while enforcing jev-guard for command safety and git-jev for pre-commit diff screening
Scanned 9/22/2026
Install to Claude Code
npx -y skills add AkashPriyadarshii/jev-superpowers --skill jev-executing-plans --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Jev Executing Plans?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/akashpriyadarshii-jev-executing-plans)More formats (shields.io, HTML) on the badges page.
---
name: jev-executing-plans
description: Use when executing a written implementation plan - guides step-by-step TDD execution while enforcing jev-guard for command safety and git-jev for pre-commit diff screening
---
# Jev Executing Plans: Safe & Verified Execution
Follows the disciplined execution loop from `superpowers:executing-plans`, but inserts deterministic System One safety gates before dangerous actions and commits.
## The Execution Loop
### 1. Load & Critical Review
- Read the plan file in `docs/superpowers/plans/`.
- Verify worktree isolation if applicable (`using-git-worktrees`).
- Create a clear checklist of tasks.
### 2. Command Safety Gate (`jev-guard`)
Before executing any potentially destructive shell command (mass deletion, recursive file changes, branch manipulation, schema migrations):
You MUST screen the command through `jev-guard`:
```bash
jev-guard "<command-to-execute>"
```
If `jev-guard` flags the command as destructive or unsafe, STOP and request explicit human confirmation.
### 3. Pre-Commit Reflex Gate (`git-jev`)
Before completing any task commit or git push:
1. Stage your changes: `git add <files>`
2. Screen the diff through `git-jev`:
```bash
git jev check
```
3. Verify that:
- No unredacted secrets or private tokens are leaked ($p < 0.10$).
- No destructive payloads or backdoor patterns exist ($p < 0.10$).
4. Commit only after `[PASS]` is reported.
### 4. Stop on Blockers
If a test fails unexpectedly or an instruction is ambiguous:
- STOP immediately.
- Do NOT force changes or attempt blind symptom fixes.
- Switch to `jev-systematic-debugging`.
## Failure Modes
See docs/CONFIDENCE.md for thresholds. When the gate tool is missing, the key is invalid, the registry is offline, or confidence falls below the Stop band: STOP, state which input failed, and never degrade to unverified guessing silently.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!